This is the very first post of the blog section! First of all, I want to thank those people who encouraged us to start the blog section with this specific topic.
The final objective of this writing is giving a set of tips on how a company should start building a solid AI strategy. The target readers for this lecture are the people whose role is related to leading the development and the execution of long-term strategies for their companies.
AI has endless advantages in today's world, and we will address that in future posts. But mainly, the vast majority of AI initiatives can be grouped into:
Horizontal applications, which are those that can be shared among many business areas, such as customer segmentation, churn rate detection, or recommendation algorithms.
Vertical applications, which refers to company-specific applications, such as detecting cancer with medical image analysis or optimal store location.
However, one of the most important and trendy aspects of AI is that it may help your company surpass its competitors when executed correctly. You can find many documents and sources of information across the internet of why AI is a good investment and why you should invest now, but again, this is not the purpose of this reading.
While many companies those days are producing an AI transformation, many of them fail due to the lack of strategy while implementing AI projects. In my opinion, half part of the success in AI projects belongs to the technical abilities of the team, and the other half comes from expectation management, social engineering and many other abilities related to the art of strategy. Those two parts are complementary, dependent, symbiotic.
Once you have decided to give AI the chance to help your business grow, you start wondering which will be the cost in terms of time and resources. With this post, I have tried to summarize our experience with the AI transformation of companies to help you at this precise step.
Moving on to the desired topic, I have pointed out a list of the main ideas that you have to keep in mind while moving your company to an AI-driven company.
Regarding pilot projects, we recommend starting at least two pilot projects at the same time. Your first projects should be a quick win to gain momentum with your AI strategy. Quick means between 8 to twelve months with a team that ranges from five to 15 people. And gaining momentum means having stronger support from the board when it comes to deciding about AI initiatives. Being that said, you should have the first hints on which can be an estimation of the initial cost of your AI journey.
Is important to wisely choose which problems one should solve at first to make the best possible impact internally and externally. I recommend you to build vertical applications instead of horizontal applications because big companies with long AI experience tend to easily outperform rookies in those projects due to the massive amount of resources that they can invest. Moreover, is less probable that big companies invest their resources in your specific use case.
The three general ideas that you should have in mind to find attractive pilot project ideas are; reducing costs, increasing revenue and opening new business lines. Avoid choosing a project just because you have a big dataset related to "X". Choose your projects based on strategy and the final added value.
One of the key factors to make your company succeed in the AI race is to have an in-house AI team. This will deliver value in many different ways such as creating IP, being known as an AI-enabled company highering your valuation and potential, and opening business lines among many others.
In my particular opinion, it is essential to have an external AI partner that helps your company during this trip of building the in-house AI team and developing the initial prototypes of the pilot projects if you want to speed up the process. Furthermore, an external AI partner is also important to deliver the fine-tuned AI expertise to your future team.
As you might have noticed while reading this text, it is very different than a Machine Learning book or a Deep Learning research paper. The board training in AI should be very different than technical training as well. The objectives are the same but the approaching to the final solution and the interests are distinct. Thus, the training process should be adapted to the specific needs of the board members.
The rest of the company members, which do not belong either to the AI team or to the board but are involved in the usage of an AI-based product, should also receive adapted AI training so they can easily understand how to use the internal products. Furthermore, this is important to create a cross-functional AI team in the future.
Correct design and execution of your AI strategy are what will bring your company to an AI-enabled company. The AI strategy should be planned in the long term (2 years is a good frame) and ideally, following the hints that you can find while reading this post.
Those questions above are two more hints for your AI strategy.
It is very important to communicate the AI projects progress to the board, to the rest of the company, and even externally. This will uphold the team's morale and will help directors or investors to allocate more resources for future AI projects.
Those hints given above are suited for companies that can spend their resources and time in the AI journey without much effort. Startups environment is different because they tend to have financial problems during the first stages of their lives, and it can be difficult to adjust the resources and deadlines to this kind of companies. However, as this is an entire topic to address and we have something to add to the discussion, we will expand the information regarding AI strategy and startups in the following posts.
This post is based on an article written by Professor Andrew Ng, which is one of the fathers of Deep Learning and has played a key role in the AI strategy of Google and Baidu.